Question

我正在实施Vitaly的pg-promise性能模式，建议here和there。

这是我的代码：

for (var i=0;i<chunkedData.length;i++){

    var insertData = chunkedData[i].map(function (d) {
        return {
            application_id: d.application_id,
            country_id: d.country_id,
            collection_id: collectionId
        };
    });

    // Would need to make a loop here, and thus turning the result into an array
    var updateData = {
        application_id: chunkedData[i][j].application_id,
        country_id: chunkedData[i][j].country_id,
        collection_id: collectionId
    };

      var query = h.insert(insertData, cs) +
          " ON CONFLICT ON CONSTRAINT application_average_ranking_application_id_country_id_colle_key DO UPDATE SET " +
          h.sets(updateData, cs);

    db.none(query)
        .then(data => {
            console.log('success');
        })
        .catch(error=> {
            console.log('insert error : ' + error);
        });
}

我的问题是insertData是一个对象数组，而库insert helper使用该数组构建一个插入请求，如pg-promise API中所述。而updateData必须是一个简单的对象。

我希望在以下时间：

ON CONFLICT ON CONSTRAINT constraintName DO UPDATE

触发

，更新值与'insertData'数组中的相应对象匹配。

我如何解决这个问题？

我试图把所有东西放在一个循环中，但它会像疯了一样泄漏内存，好吧，我失去了模式的好处......

编辑：

我希望我的查询相当于：

 var inserts = data.map(entry => {
    return t.none(" INSERT INTO application_average_ranking (application_id,country_id,collection_id) VALUES ($1,$2,$3)" +
                 " ON CONFLICT ON CONSTRAINT application_average_ranking_application_id_country_id_colle_key" +
                 " DO UPDATE SET country_id=$2,collection_id=$3",
                 [entry.application_id,entry.country_id,collectionId]
    );
});

在调用Update时，参数引用最初建议插入的值。

Answer 1

您的任务需要一个静态SQL来实现这种逻辑，方法是使用EXCLUDED作为表引用，其行由于冲突而排除：

var sqlConflict = " ON CONFLICT ON CONSTRAINT" +
    " application_average_ranking_application_id_country_id_colle_key" +
    " DO UPDATE SET application_id = excluded.application_id" +
    " country_id = excluded.country_id, collection_id = excluded.collection_id";

var insertData = chunkedData.map(function (d) {
    return {
        application_id: d.application_id,
        country_id: d.country_id,
        collection_id: collectionId
    };
});

var query = h.insert(insertData, cs) + sqlConflict;

db.none(query)
    .then(data => {
        console.log('success');
    })
    .catch(error=> {
        console.log('insert error : ' + error);
    });

<强>更新

如果排除字段的静态列表太长并且您想要简化它，您可以始终依赖helpers方法的灵活性：

// or pull them from an object using `Object.keys(obj)`:
var cols = ['application_id', 'country_id', 'collection_id'];

var sets = pgp.helpers.sets({}, cols.map(c=> ({
    name: c, mod: '^', def: 'excluded.' + pgp.as.name(c)
})));

console.log(sets);
//=> "application_id"=excluded."application_id","country_id"=excluded."country_id",
//   "collection_id"=excluded."collection_id"

// or its simple JavaScript equivalent:
var sets = cols.map(c=> {
    var name = pgp.as.name(c);
    return name + '=excluded.' + name;
}).join();

<强>更新

对于库的7.3.0及更高版本，您应该使用方法assignColumns来生成所有排除的集合，如下所示：

cs.assignColumns({from: 'EXCLUDED'})
//=> "application_id"=EXCLUDED."application_id","country_id"=EXCLUDED."country_id","collection_id"=EXCLUDED."collection_id"

或者，如果您想跳过application_id，则可以执行以下操作：

cs.assignColumns({from: 'EXCLUDED', skip: 'application_id'})
//=> "country_id"=EXCLUDED."country_id","collection_id"=EXCLUDED."collection_id"

请参阅ColumnSet.assignColumns

Answer 2

请勿使用h.sets()。只需自己写conflict_action即可。手册说

ON CONFLICT DO UPDATE中的SET和WHERE子句可以使用表的名称（或别名）访问现有行，并使用特殊排除表访问建议插入的行。

Postgres - Insert

Pg-promise性能提升：具有多个更新参数的多个插入

2 个答案: